Artificial intelligence is now being applied in many fields, so it is only natural to consider its applications in the markets. After all, has the financial sector—banks, insurance companies, intermediaries and others—not long been one of the world's leading "consumers" of computing resources? One might therefore expect it to be among the pioneers in the field.
Yet that is far from the case. Although the leaderboards for machine learning competitions no longer feature universities alone, the companies listed are Google and Facebook, not Goldman Sachs or JP Morgan—even though the latter are two of the sector's most innovative firms.
What explains this anomaly?
-
Protection against risk
-----------------------
To understand why the financial sector is not among the leaders in AI, we must first recognize that it is one of the most highly regulated sectors. As successive crises have shown, it encompasses activities in which the slightest malfunction can affect a large part of the global economy and thereby worsen the living conditions of many households. One of the central aims of regulation is to ensure that the sector's principal institutions—known as systemically important institutions—act primarily as intermediaries for financial risk. This means that instead of taking risks themselves, they divide them up and circulate them, enabling participants outside the financial sector to protect themselves against all kinds of risk, provided they use these institutions.
The standard example—a little simplistic, but helpful—is that of a farmer who wants insurance against bad weather and the owner of a nearby hydroelectric power station who wants protection against fine weather. A financial intermediary can then arrange two "opposing" contracts without anyone genuinely taking on risk. In this simplified textbook case, we can imagine that none of the three parties remains financially exposed to fluctuations in the weather. These are replaced by deterministic cash flows: insurance premiums and margins.
